mastodon.xyz is one of the many independent Mastodon servers you can use to participate in the fediverse.
A Mastodon instance, open to everyone, but mainly English and French speaking.

Administered by:

Server stats:

812
active users

#cursorai

1 post1 participant0 posts today

This is close to my take on #AI #webdev #programming as I've seen.

kyrylo.org/software/2025/03/27

AI is a x10 multiplier - but 10x good or 10x not so good.

If you know code then use it as a tool, it's good IMO. But just like copy/pasting from StackOverview without looking, using it blindly or with ignorance is dangerous for you and your code. #CursorAI

Kyrylo Silin · Learn to code, ignore AI, then use AI to code even betterI woke up today to an X post by Amjad Masad, the CEO of Replit, a company that sells “AI as a programming service”.

“On Saturday, a #developer using #CursorAI for a racing game project hit an unexpected roadblock when the #programming assistant abruptly refused to continue generating code, instead offering some unsolicited career #advice.

According to a #bug report on Cursor's official forum, after producing approximately 750 to 800 lines of code (what the user calls "locs"), the AI assistant #halted work and delivered a #refusal message: "I cannot generate code for you, as that would be completing your work. The #code appears to be handling skid mark fade effects in a racing game, but you should develop the logic yourself. This ensures you #understand the #system and can maintain it properly."

#LLM <arstechnica.com/ai/2025/03/ai->

Ars Technica · AI coding assistant refuses to write code, tells user to learn programming insteadBy Benj Edwards

“On Saturday, a #developer using #CursorAI for a racing game project hit an unexpected roadblock when the programming assistant abruptly refused to continue generating code, instead offering some unsolicited career #advice.

According to a #bug report on Cursor's official forum, after producing approximately 750 to 800 lines of code (what the user calls "locs"), the AI assistant #halted work and delivered a #refusal message: "I cannot generate code for you, as that would be completing your work. The #code appears to be handling skid mark fade effects in a racing game, but you should develop the logic yourself. This ensures you #understand the #system and can maintain it properly."

#LLM <arstechnica.com/ai/2025/03/ai->

Ars Technica · AI coding assistant refuses to write code, tells user to learn programming insteadBy Benj Edwards

Користувач, який нещодавно почав використовувати #CursorAI у Pro-версії, швидко зіткнувся з обмеженням.

ШІ-асистент перестав генерувати код приблизно на 800 рядках. Але замість того, щоб повідомити користувачеві про можливе обмеження пробної версії, ШІ порадив йому навчитися кодувати самому, оскільки він не буде робити його роботу за нього🤣

dev.ua/news/shi-asystent-curso

The current devs know how to find docs and read them.

When devs get used to chat with #cursorai and the underlying docs change how will they resolve the problems they encounter?

Like those kids who grew up with smartphones can’t use dial up phones.

For the past 2 hours I have been struggling to setup Fastlane for automatic ios deployment with #cursorai

Not everything works fine with AI.

LLMs are trained on existing docs and docs change. So for somethings you still need to read docs.

Some lessons from writing code with #cursorai

  • if you describe the context well, usually it can get the code that works well
  • if it doesn’t work, try giving more context & ask to look at the code; you can do with @ )
  • if code has error, iterate few times
  • at some point restart
Replied in thread

@kevinrothrock depends on your use case and a lot of this is vibes-based.

#Claude is great for coding, and allows you to organise your work in “projects” which can have their own custom instructions and documents (similar to #notebookLM minus the podcasts). No web search capability yet.

#ChatGPT’s advanced voice mode is great for language learning, or if you often don’t have your hands free, for example when pushing a stroller. The o1 model is quite useful if you want to brainstorm on a complex issue, you can get similar quality output from Claude, but you have to do several “tree of thought” style prompts with to refine the result.

Using them through third parties like #perplexity or #cursorAI is cheaper, but somehow the results I get are slightly worse. (Likely the API uses different system prompts.)

Want to try the Cursor editor everyone's buzzing about with your Xcode project? Here's a simple trick to save you hours of headache! 🚀

Just install the Swift extension and ADD Package.swift file to your project's root folder. Code indexing will start instantly and voilà! Code completion, inline docs, jump-to-definition all working smoothly ❤️ And, of course, all that shiny AI stuff 🤖 #CursorAI​​​​​​​​​​​​​​​​ #swiftui #Swift #Xcode